KR20200015678A - Apparatus and method for coding/decoding image selectivly using descrete cosine/sine transtorm - Google Patents

Abstract

The present invention relates to an image encoding technique and an image decoding technique. According to the present invention, an image encoding apparatus comprises: a first transform part for transforming a target image into a first transform coefficient by using a discrete cosine transform (DCT) technique; a second transform part for transforming the target image into a second transform coefficient by using a discrete sine transform (DST) technique; and a transform technique determination part selecting any one of the first transform coefficient and the second transform coefficient as a transform coefficient for the target image.

Description

Coding / decoding apparatus and method for selectively using discrete cosine transform / discrete sine transform {APPARATUS AND METHOD FOR CODING / DECODING IMAGE SELECTIVLY USING DESCRETE COSINE / SINE TRANSTORM}

The present invention relates to an image encoding technique and an image decoding technique.

In general, a video encoding method includes intra coding that performs intra-picture encoding, such as an intra frame (I), and inter coding that performs inter-picture encoding, such as a P (predictive) frame or a B (bidirectional) frame. Can be distinguished.

Motion estimation in video compression standards such as H.263, MPEG-4 Part 2, and H.264 / MPEG-4 AVC is performed in units of blocks. That is, motion estimation is performed in units of macroblocks, or motion estimation is performed in units of subblocks obtained by dividing or subdividing macroblocks. Motion estimation is performed to reduce bitrate by removing temporal redundancy in video encoding. In particular, H.264 / MPEG-4 AVC has high coding efficiency by using variable block-based motion estimation of various sizes.

Prediction of a motion vector is performed by referring to a past image on the basis of a time axis or by referring to both a past image and a future image. An image referred to for encoding or decoding a current frame is called a reference image. Since H.264 / MPEG-4 AVC supports multiple reference frames, coding is larger than when using the previous frame as the reference picture by selecting the block of the frame having the most overlapping with the current block as the reference picture. Efficiency can be obtained. In addition, to select the most optimal mode among all possible coding modes such as variable block mode, spatial prediction mode (Intra 16x16, Intra 8x8, Intra 4x4), SKIP mode (P_SKIP mode and B_SKIP mode), DIRECT mode, etc. The rate-distortion optimization technology is used to further improve the coding efficiency of H.264 / MPEG-4 AVC.

According to the H.264 / MPEG-4 AVC standard, which is designed to encode and decode video data, the conversion method is a method for reducing the spatial correlation of residual signals in a block and increasing the compression ratio of energy after performing intra / intra picture prediction. The quantizer is used to increase the compression ratio by further reducing the energy of the transformed coefficient after the use of the transform method.

An object of the present invention is to increase the compression rate of an image block when generating quantized transform coefficients through transform and quantization after performing inter- / intra-prediction of a block (macroblock) of a predetermined size.

In order to achieve the above object and to solve the problems of the prior art, the present invention provides a first transform unit for generating a first transform coefficient using a Discrete Cosine Transform (DCT) technique, the target image A second transform unit generating a second transform coefficient using a Discrete Sine Transform (DST) technique, and a transform coefficient of the first transform coefficient and the second transform coefficient for the target image Provided is a video encoding apparatus comprising a conversion scheme determination unit for selecting.

According to one aspect of the present invention, a transform scheme determination unit for determining a transform scheme for a transform coefficient, when the transform scheme is a discrete cosine transform, a first method for generating a decoded image of the transform coefficient using a discrete cosine inverse transform technique When the inverse transform unit and the transform scheme is a discrete sine transform, a video decoding apparatus including a second inverse transform unit for generating a decoded image of the transform coefficients using a discrete sine inverse transform scheme is provided.

According to another aspect of the present invention, a first transform unit for performing a discrete cosine transform on the target image to generate a first transform coefficient, a second transform coefficient for performing a discrete sine transform on the target image to generate a second transform coefficient And a second transform unit and a transform scheme determiner configured to select one of the first transform coefficient and the second transform coefficient as a transform coefficient for the target image.

According to the present invention, the compression rate of an image block can be increased when generating quantized transform coefficients through transform and quantization after performing inter- / intra-prediction of a block (macroblock) of a predetermined size.

1 is a block diagram showing an embodiment of an image encoding apparatus to which the present invention is applied.
2 is a block diagram illustrating a structure of an image encoding apparatus according to an embodiment of the present invention.
3 is a diagram illustrating a concept of integer sine transform according to an embodiment of the present invention.
4 is a diagram illustrating a value of a multiplication factor in a quantization process for a 4 × 4 image.
FIG. 5 is a diagram illustrating a multiplication factor value in a quantization process for an 8x8 image.
FIG. 6 is a diagram illustrating a value of a scaling factor in a dequantization process for a 4 × 4 sized image.
FIG. 7 is a diagram illustrating a value of a scaling factor in an inverse quantization process for an 8x8 image.
8 illustrates a concept of an integer sine inverse transform according to an embodiment of the present invention.
9 is a diagram illustrating the concept of a coefficient threshold method according to an embodiment of the present invention.
10 illustrates a concept of storing transform scheme information in consideration of a coded block pattern according to an embodiment of the present invention.
FIG. 11 is a diagram illustrating an embodiment of the present invention for storing information on a conversion scheme in a macro block header. FIG.
FIG. 12 is a diagram illustrating an embodiment of the present invention for storing in a slice header whether or not the present invention is applied.
13 is a block diagram showing the structure of a decoder according to the present invention.
14 is a block diagram illustrating a structure of an image encoding apparatus according to another embodiment of the present invention.

Hereinafter, with reference to the accompanying drawings will be described an embodiment of the present invention;

1 is a block diagram showing an embodiment of an image encoding apparatus to which the present invention is applied.

The image encoding apparatus includes a transform and quantization unit 110, an entropy coding unit 120, a coding control unit 130, an inverse quantization and inverse transform unit 140, a loop filter 150, a reference image storage unit 160, and motion estimation. / Compensator 170 and intra prediction / compensator 180 are provided.

In general, the encoding apparatus includes an encoding process and a decoding process, and the decoding apparatus includes a decoding process. The decoding process of the decoding apparatus is the same as the decoding process of the encoding apparatus. Therefore, the following description focuses on the encoding apparatus.

The motion estimation / compensation unit 170 and the intra prediction / compensation unit 180 generate a prediction image for the input image. The motion estimator / compensator 170 and the intra predictor / compensator 180 generate a predicted image of the input image with reference to the reference image of the input image. The reference image storage unit 160 may store a reference image for the input image. The difference between the input image and the predicted image may be referred to as a residual image. According to an embodiment of the present invention, the image encoding apparatus may improve encoding efficiency than when encoding the residual image by encoding the residual image. That is, the residual image may be used as the target image of the image encoding device.

The target image is input to the transform and quantization unit 110. The transform and quantization unit 110 generates a first transform coefficient by performing a discrete cosine transform (DCT), quantization, and coefficient threshold on the target image. The transform and quantization unit 110 also performs Discrete Cosine Transform (DST) quantization and coefficient thresholds to generate second transform coefficients. The entropy encoder 120 generates a bitstream by performing entropy coding on the discrete transformed cosine transformed first transform coefficient or the discrete sine transformed second transform coefficient. In this case, the coding controller 130 performs inverse quantization and discrete cosine inverse transform on the first transform coefficient, and inverse quantization and discrete sine inverse transform on the second transform coefficient to determine an optimal transform technique. The coding controller 130 transmits the determined transform scheme to the transform and quantization unit 110.

Referring to the decoding process 190, the inverse quantization and inverse transform unit 140 performs inverse quantization and discrete cosine inverse transformation on the first transform coefficient to generate a decoded image of the first transform coefficient, and performs the second transform coefficient on the second transform coefficient. Inverse quantization and discrete sine inverse transform are performed to generate a decoded image for the second transform coefficients. The loop filter 150 performs low pass filtering on each decoded image to smooth block boundaries. The reference image storage unit 160 stores the decoded image having the smooth block boundary as the reference image.

In the inter mode, the motion compensator 170 performs motion compensation by comparing the stored reference image with the input image. In the intra mode, the intra prediction / compensator 180 performs intra compensation without passing through the loop filter 150 and the reference image storage unit 160.

2 is a block diagram illustrating a structure of an image encoding apparatus according to an embodiment of the present invention.

The image predictor 210 generates a predicted image of the input image by referring to the reference image of the input image. The image predictor 210 calculates a difference between the input image and the predicted image as the target image.

The first transform unit 220 converts the target image into first transform coefficients using a Discrete Cosine Transform (DCT) technique. The first transform unit 220 may include a discrete cosine transform unit 221 and a quantization unit 222.

The discrete cosine transforming unit 221 generates a first image coefficient by performing discrete cosine transforming of the target image. The quantization unit 222 generates a first transform coefficient by quantizing the first image coefficient.

The second transform unit 230 converts the target image into second transform coefficients using a Discrete Sine Transform (DST) technique. The second transform unit 230 may include a discrete sine transform unit 231 and a quantization unit 232.

The operations of the discrete cosine transform unit 221 and the quantization unit 222 included in the first transform unit 220 are performed by the discrete sine transform unit 231 and the quantization unit 232 included in the second transform unit 230. Since it is similar to the operation, the operation of the second converter 230 will be described in detail below.

The discrete sine transform unit 231 performs discrete sine transform of the target image to generate a second image coefficient. The quantization unit 232 quantizes the second image coefficient to generate a second transform coefficient.

According to an embodiment, the discrete sine transform unit 231 may perform discrete sine conversion of the target image according to Equation 1 below.

[Equation 1]

는 이산 정현 변환될 목적 영상을 나타내고,

는 이산 정현 변환된 제2 영상 계수를 나타낸다. 또한

은 이산 정현 변환의 단위 크기를 나타낸다.here,

Represents the target image to be discrete sine transformed,

Represents a discrete sine transformed second image coefficient. Also

Denotes the unit size of the discrete sine transform.

When the target image is 4x4 in size, the discrete sine transform of Equation 1 may be expressed as a matrix as shown in Equation 2 below.

[Equation 2]

는

의 각 행에 대한 이산 정현 변환 행렬을 나타내며,

는 목적 영상,

는 역양자화된 제2 영상 계수를 나타낸다. 또한 행렬 내의 원소

와

는 각각 상수

와

를 나타낸다.

Is

Represents a discrete sine transformation matrix for each row of,

Purpose video,

Denotes the dequantized second image coefficient. Also elements within a matrix

Wow

Are constants

Wow

Indicates.

와

는 정수가 아니므로 실제 계산하는 경우 계산 과정이 매우 복잡하다. 일실시예에 따르면 이산 정현 변환부(231)는 이산 정현 변환 행렬로부터 유도된 정수 정현 변환 행렬을 이용하여 목적 영상을 정수 정현 변환할 수 있다. 정수 정현 변환은 하기 수학식 3과 같이 표현할 수 있다.Elemental elements in the matrix

Wow

Since is not an integer, the calculation process is very complicated when it is actually calculated. According to an embodiment, the discrete sine transform unit 231 may perform integer sine transform on the target image by using an integer sine transform matrix derived from the discrete sine transform matrix. The integer sine transform can be expressed as in Equation 3 below.

[Equation 3]

3 is a diagram illustrating a concept of integer sine transform according to an embodiment of the present invention.

If the target image is 4x4 in size, the integer sine transform in Equation 3 may be simply performed according to the embodiment shown in FIG.

When the target image is 8x8 in size, the discrete sine transform of Equation 1 may be expressed as a matrix as shown in Equation 4 below.

[Equation 4]

The quantization unit 232 quantizes the second image coefficient to generate a second transform coefficient. According to an embodiment, when the target image has a size of 4x4, the quantization unit 232 may quantize the second image coefficient by using Equation 5 or Equation 6 below.

[Equation 5]

[Equation 6]

는 4x4 행렬 (i,j) 위치에서 정수 정현 변환된 제2 영상 계수를 나타내고,

는 4x4 행렬 (i,j) 위치에서의 제2 변환 계수가 양자화되어 생성된 제2 변환 계수를 나타낸다.

는

의 값을 나타내며,

는 양자화 매개변수(Quantization Parameter)를 나타낸다.

Represents an integer sine transformed second image coefficient at position 4x4 matrix (i, j),

Denotes a second transform coefficient generated by quantizing the second transform coefficient at the 4 × 4 matrix (i, j) position.

Is

Represents the value of,

Denotes a quantization parameter.

는 곱셈 인자(Multiplication Factor)를 나타낸다.

은 양자화기의 스탭 크기(Step Size),

는 반올림 함수를 나타낸다.

는 양자화 과정에서의 데드 존(dead zone)을 조정하는 역할을 하며, 0에서 1/2 사이의 값으로 선택된다.

는 부호 함수(sign function)이다.

Denotes a multiplication factor.

Is the step size of the quantizer,

Represents a rounding function.

Is used to adjust the dead zone in the quantization process, and is selected as a value between 0 and 1/2.

Is a sign function.

4 is a diagram illustrating a value of a multiplication factor in a quantization process for a 4 × 4 image.

According to an embodiment, when the target image is 8x8 in size, the quantization unit 232 may generate a second transform coefficient by quantizing the second image coefficient using Equation 7 below.

[Equation 7]

FIG. 5 is a diagram illustrating a multiplication factor value in a quantization process for an 8x8 image.

The transformation technique determiner 260 selects any one of the first transformation coefficient and the second transformation coefficient as a transformation coefficient for the target image.

According to an embodiment, the first transform unit 220 generates a first decoded image of the target image based on the first transform coefficients, and the second transform unit 230 generates a first decoded image based on the second transform coefficients. A second decoded image may be generated. The conversion technique determiner 260 may select a converted image for the target image by comparing the target image, the first decoded image, and the second decoded image.

According to another embodiment, the conversion technique determiner 260 further considers the bit amount of the first decoded image, the bit rate of the first decoded image, the bit amount of the second decoded image, and the bit rate of the second decoded image. A converted image for the target image may be selected. That is, the transform scheme determiner may select a transform coefficient corresponding to a decoded image having a smaller bit rate or a lower bit rate among the first decoded image and the second decoded image as a transform coefficient for the target image.

According to another embodiment, the conversion technique determiner 260 may select the converted image for the target image by further considering the distortion of the first decoded image and the distortion of the second decoded image. As an example, the transform scheme determiner 260 may select a transform coefficient for the target image in consideration of the distortion and the bit rate of the decoded image.

The first transform unit 220 may include a discrete cosine transform coefficient threshold 223, an inverse quantizer 224, and a discrete cosine inverse transform unit 225 to generate a first decoded image.

The discrete cosine transform coefficient threshold unit 223 performs a coefficient threshold operation on the first transform coefficient. The inverse quantization unit 224 inverse quantizes the first transform coefficient having a coefficient threshold operation, and the discrete cosine inverse transform unit 224 performs discrete inverse transform on the inverse quantized first transform coefficient to perform a first decoded image on the target image. Create The transform technique determiner 260 selects a transform coefficient based on the first decoded image.

The second transform unit 230 may include a discrete sine transform coefficient threshold 233, an inverse quantizer 234, and a discrete sine inverse transform unit 225 to generate a second decoded image.

The operations of the discrete cosine transform coefficient threshold unit 223, the inverse quantization unit 224, and the discrete cosine inverse transform unit 225 included in the first transform unit 220 are performed by the discrete sine transform included in the second transform unit 230. Since the operations of the coefficient threshold unit 233, the inverse quantization unit 234, and the discrete sine inverse transform unit 225 are similar, the operation of the second transform unit 230 will be described in detail below.

The discrete sinusoidal transform coefficient threshold 233 performs a coefficient threshold operation on the quantized second transform coefficients.

The coefficient threshold of the discrete sine transform coefficient threshold 233 will be described in detail with reference to FIG. 9.

The quantized second transform coefficients are represented by 3, 2, 1, and 0 on the left side of FIG. 4 for coefficients occurring in each 4x4 frequency domain through a coefficient thresholding method as shown in FIG. 9 below. ), Calculate the total sum of costs in 8x8 block units, and if the cost is not higher than a certain threshold, replace all the transformed coefficients of the corresponding 8x8 block with 0, and coded block pattern for the block. ) To 0. The cost described above applies only when the absolute value of the transform coefficient is 1 in the frequency domain.

At this time, if the absolute value of the coefficient in any one frequency domain is greater than 1, the cost for the corresponding transform coefficient is a 4x4 block and 8x8 blocks. use.

In addition, if the sum of costs of four 8x8 block units in macroblock units is not higher than a specific threshold, all coefficients of the corresponding macroblocks are replaced with 0, and a coded block pattern for each 8x8 block is included. Determine 0 as 0.

The H.264 / AVC reference model JM (Joint Model) and the ITU-T VCEG KTA reference model use a fixed constant of 4 in 8x8 block units and a threshold of 5 in macroblock units. use.

In FIG. 9A, since the total cost sum of the transform coefficients of the 8 × 8 block is 5, all coefficient values are not replaced with zero. Since the total cost sum of the transform coefficients of the 8x8 block in FIG. 9B becomes zero, the total coefficients in FIG. 9B are replaced by zero.

The inverse quantization unit 234 inversely quantizes the second transform coefficient on which the coefficient threshold operation is performed. According to an embodiment, when the target image has a 4x4 size, the inverse quantization unit 234 may inverse quantize the second transform coefficient using Equation (8).

[Equation 8]

는 역양자화된 이후의 제2 변환 계수이고,

는 크기 조정 인자(Scaling Factor)이다.

Is the second transform coefficient after dequantization,

Is a scaling factor.

의 값을 나타낸 도면이다.6 shows scaling factors in the inverse quantization process for a 4x4 sized image.

Shows the value of.

According to an embodiment, when the target image is 8x8 in size, the inverse quantization unit 234 may inverse quantize the second transform coefficient on which the coefficient threshold operation is performed, using Equation (9).

[Equation 9]

의 값을 나타낸 도면이다.7 shows the scaling factor in the inverse quantization process for an 8x8 image

Shows the value of.

The discrete sinusoidal inverse transform unit 235 generates a second decoded image of the target image by performing discrete inverse transform on the inverse quantized second transform coefficients. According to an embodiment, the discrete sine inverse transform unit 235 may generate a second decoded image according to Equation 10 below.

[Equation 10]

는

의 각 행에 대한 이산 정현 역변환 행렬을 나타내며,

는 목적 영상,

는 역양자화된 변환 계수를 나타낸다. 또한 행렬 내의 원소

와

는 각각 상수

와

를 나타낸다.

Is

Represents the discrete sine inverse matrix for each row of,

Purpose video,

Denotes a dequantized transform coefficient. Also elements within a matrix

Wow

Are constants

Wow

Indicates.

와

는 정수가 아니므로 실제 계산하는 경우 계산 과정이 매우 복잡하다. 일실시예에 따르면 이산 정현 역변환부(235)는 이산 정현 변환 행렬로부터 유도된 정수 정현 역변환 행렬을 이용하여 제2 복호 영상을 정수 정현 역변환할 수 있다. 정수 정현 역변환은 하기 수학식 11과 같이 표현할 수 있다.Elemental elements in the matrix

Wow

Since is not an integer, the calculation process is very complicated when it is actually calculated. According to an embodiment, the discrete sine inverse transform unit 235 may inversely transform the second decoded image by using the integer sine inverse transformation matrix derived from the discrete sine transformation matrix. The integer sine inverse transform may be expressed as in Equation 11 below.

[Equation 11]

8 is a diagram illustrating a concept of integer sine transform according to an embodiment of the present invention.

If the target image is 4x4 in size, the integer sine inverse transform in Equation 11 may be simply performed according to the embodiment shown in FIG. 8.

은 하기 수학식 12과 같은 반올림 과정을 거쳐

로 표현될 수 있다.Second Decoded Image Generated Using Discrete Sine Inverse Transform

After the rounding process as shown in Equation 12

It can be expressed as.

[Equation 12]

은 하기 수학식 13와 같은 후-크기 조정(Post-Scaling)과정을 거쳐

로 표현된다.Meanwhile, the second decoded image reconstructed by the integer sine inverse transform

After the post-scaling process as shown in Equation 13

It is expressed as

[Equation 13]

The Lagrange multiplier calculator 240 calculates a Lagrange Multiplier for the first decoded image and the second decoded image, and the distortion rate measurer 250 based on the calculated Lagrange multiplier, respectively. The distortion rate of the decoded image is measured, and the distortion rate of the second decoded image is measured.

According to an embodiment, in the case of an I-frame or a P-frame, the Lagrange multiplier calculator 240 may calculate the Lagrange multiplier according to Equation 14 below.

[Equation 14]

의 값은 여러 영상들에 대하여 실험을 통해 결정될 수도 있으며, 목적 영상의 크기에 따라서

의 값이 적응적으로 결정될 수 있다.

The value of may be determined through experiments on several images, depending on the size of the target image.

The value of can be adaptively determined.

According to another embodiment, in the case of a B-frame, the Lagrange multiplier calculator 240 may calculate the Lagrange multiplier according to Equation 15 below.

* [Equation 15]

The rate-distortion cost calculator 250 calculates a rate-distortion cost (R-D cost) for the first decoded image and the second decoded image.

According to an embodiment, the rate-distortion cost calculator 250 may calculate the rate-distortion cost according to Equation 16 below.

[Equation 16]

는 각 복호 영상의 왜곡이고,

는 각 복호 영상의 비트량이다.

는 영상 부호화에서 사용되는 부호화 모드이다.Here, s is a target video and c is a decoded video.

Is the distortion of each decoded image,

Is the bit amount of each decoded video.

Is an encoding mode used in image encoding.

The rate-distortion cost calculation unit 260 selects an optimal mode among all possible encoding modes, such as a variable block mode used in image encoding, four spatial prediction modes, a SKIP mode (P_SKIP mode and B_SKIP mode), and a DIRECT mode. Calculate the rate-distortion cost for each mode. That is, the rate-distortion cost calculation unit 260 calculates the rate-distortion cost for each encoding mode for each reconstructed image.

The transformation technique determiner 260 may select a transformation coefficient corresponding to the reconstructed image having the minimum rate-distortion cost as the transformation coefficient for the target image. That is, the transformation technique determiner 260 may select an optimal transformation scheme by considering not only the distortion of each reconstructed image but also the bit rate required for encoding the transformation coefficient.

In addition, the conversion scheme determination unit 260 may determine a mode in which a rate-distortion cost (R-D cost) is minimum as an encoding mode. That is, the conversion technique determiner 260 may select a conversion mode having a high compression efficiency as a conversion mode for the target image.

The transformation scheme storage unit 270 may store information on a transformation scheme corresponding to the transformation coefficient of the target image. That is, the conversion technique storage unit 270 stores whether the target image is discrete cosine transformed or discrete sine transformed, and the image decoder may determine the conversion technique for the target image based on the information on the conversion technique. .

According to an embodiment, the conversion scheme storage unit 270 may display information on the conversion scheme in flag bits provided in macroblock units. For example, when the target image included in the macroblock is discrete cosine transformed, the flag bit may be determined as '0'. Alternatively, when the target image included in the macroblock is discrete sine transformed, the flag bit may be determined as '1'.

The transformation scheme storage unit 270 may store information on the transformation scheme in consideration of the coded block pattern (CBP) value of the transformation coefficient. The encoded block pattern value indicates a luminance value of each block. If the block has a transform coefficient, the encoded block pattern value is displayed as '1'. If there is no transform coefficient in a particular block, there is no need to determine the transform scheme for that block, so there is no need to store information about the transform scheme.

10 illustrates a concept of storing transform scheme information in consideration of a coded block pattern according to an embodiment of the present invention.

In FIG. 10A, an embodiment without transform coefficients is illustrated in all blocks 1011, 1012, 1013, and 1014. In this case, the decoder does not need to determine the transform scheme for the transform coefficient for each block. Therefore, the conversion scheme storage unit 270 does not store information on the conversion scheme of each block. That is, the transform scheme storage unit 270 may store information on the transform scheme for the block only when the encoded block pattern value for each block is '1'.

In FIG. 10B, an embodiment in which the second block and the third block do not have transform coefficients is illustrated. The first block 1021 and the fourth block 1024 have transform coefficients, and each transform coefficient has a discrete sine transform and a discrete cosine transform.

The transformation scheme storage unit 270 may store the transformation scheme information only for the first block 1021 and the fourth block 1041 including the transformation coefficients. Since the first block 1021 is discrete sine transformed, the information about the transform scheme may be determined as '1'. In addition, since the fourth block 1024 is discrete cosine transformed, the information on the conversion scheme may be determined as '0'.

The decoder may determine whether the information about the transform scheme of '10' indicates a transform scheme in consideration of the encoded block pattern value for each block.

In FIG. 10C, an embodiment in which only a first block 1031 has a transform coefficient and a second block 1032, a third block 1033, and a fourth block 1034 has no transform coefficient is illustrated. The transformation scheme storage unit 270 may store the transformation scheme information only for the first block 1031 having the transformation coefficient.

FIG. 10D illustrates an embodiment in which only the first block 1041 has no transform coefficient, and the second block 1042, the third block 1043, and the fourth block 1044 have the transform coefficient. The transformation scheme storage unit 270 may store the transformation scheme information only for the second block 1042, the third block 1043, and the fourth block 1044 having the transformation coefficient.

FIG. 11 is a diagram illustrating an embodiment of the present invention for storing information on a conversion scheme in a macro block header. FIG.

Referring to a portion 1110 of the macroblock header illustrated in FIG. 11, which stores information about a transform scheme, the transform scheme storage unit 270 may convert the transform scheme when the encoded block pattern is greater than '0'. Information about (alternative transform flag) can be stored.

FIG. 12 is a diagram illustrating an embodiment of the present invention for storing in a slice header whether or not the present invention is applied.

According to an embodiment, the image encoding apparatus may encode the target image by applying the present invention, but according to another embodiment, the image encoding apparatus may not apply the present invention. The transform scheme storage unit 270 may store 1210 whether the present invention is applied using the flag bits of the slice header, and the decoder may determine whether the present invention is applied to the transform coefficients using the flag bits. .

13 is a block diagram showing the structure of a decoder according to the present invention. The decoder according to the present invention includes a transform technique determination unit 1310, a first inverse transform unit 1320, and a second inverse transform unit 1330.

The transformation technique determination unit 1310 may determine whether the present invention is applied to the transformation coefficient. According to an embodiment, when the image encoding apparatus encodes the target image by applying the present invention, the image encoding apparatus may express whether the present invention is applied to the transform coefficient by using flag bits of the slice header. The transformation technique determination unit may determine whether the present invention is applied to the transformation coefficient by using flag bits of the slice header.

In addition, the transform scheme determination unit 1310 determines a transform scheme for the transform coefficients. When the present invention is applied to a transform coefficient, the transform coefficient is discrete sine transform or discrete cosine transform. The image encoding apparatus may display a transform scheme applied to a transform coefficient by using flag bits of a macroblock. In other words, the flag bits of the macro block indicate conversion scheme information. The transformation technique determination unit 1310 may determine the transformation scheme for the transformation coefficient by using flag bits of the macro block.

According to an embodiment, the flag bits of the macro block may be stored in consideration of the encoded block pattern value for the transform coefficient. That is, the image encoding apparatus may store the flag bits in the macro block only when the encoded block pattern value is not '0', and may not store the flag bits in the macro block when the encoded block pattern value is '0'. have.

When the discrete sine transform is applied to the transform coefficient, the first inverse transform unit 1320 generates a decoded image of the transform coefficient by using the discrete sine inverse transform technique. The first inverse transform unit 1320 may include an inverse quantization unit 1321 and a discrete sine inverse transform unit 1322.

The inverse quantization unit 1321 dequantizes the quantized transform coefficients. Since inverse quantization has been described in detail with reference to FIG. 2, the following detailed description will be omitted.

The discrete sine inverse transform unit 1322 generates a decoded image by performing inverse transform on the inverse quantized transform coefficients. The discrete sine inverse transform has been described in detail with reference to FIG. 2, and thus, the detailed description thereof will be omitted.

When the discrete cosine transform is applied to the transform coefficient, the second inverse transform unit 1330 generates a decoded image of the transform coefficient by using the discrete cosine inverse transform technique. The second inverse transform unit 1330 may include an inverse quantization unit 1331 and a discrete cosine inverse transform unit 1332.

The inverse quantization unit 1331 inversely quantizes the quantized transform coefficients. Since inverse quantization has been described in detail with reference to FIG. 2, the following detailed description will be omitted.

The discrete cosine inverse transform unit 1332 generates a decoded image by inversely transforming the inverse quantized transform coefficients.

According to an embodiment, the decoded image generated by the first inverse transform unit 1320 or the second inverse transform unit 1330 is not an input image input to the image encoding apparatus, but a residual image that is a difference between the input image and the prediction image. ) Corresponds to

The image reconstructor 1340 generates a predicted image of the transform coefficients by referring to the reference image of the transform coefficients. The image reconstructor 1340 generates a reconstructed image of the transform coefficients based on the predicted image and the decoded image. The reconstructed image corresponds to the input image input to the image encoding apparatus.

14 is a block diagram illustrating a structure of an image encoding apparatus according to another embodiment of the present invention. The image encoding apparatus may include an image predictor 1410, a first transformer 1420, a second transformer 1430, a distortion measurer 1440, a transform scheme determiner 1450, and a transform scheme storage unit 1460. It may include.

The image predictor 1410 generates a predicted image of the input image by referring to the reference image of the input image. Also, the image predictor 1410 calculates a difference between the input image and the predicted image as the target image. That is, the target image encoded by the first transform unit 1420 or the second transform unit 1430 is a residual image of the input image.

The first transform unit 1420 performs a discrete cosine transform on the target image to generate a first transform coefficient. In addition, the first transform unit 1420 performs a discrete cosine inverse transform on the first transform coefficient to generate a decoded image of the first transform coefficient.

The second transform unit 1430 generates a second transform coefficient by performing discrete sine transform on the target image. In addition, the second transform unit 1430 generates a decoded image of the second transform coefficient by performing a discrete sine inverse transform on the first transform coefficient.

The distortion measurer 1440 measures the distortion rate of the decoded image with respect to the first transform coefficient by comparing the target image with the decoded image with respect to the first transform coefficient. In addition, the distortion measurer 1440 measures the distortion rate of the decoded image of the second transform coefficient by comparing the decoded image of the target image with the second transform coefficient.

The transform technique determiner 1450 selects a transform coefficient for the target image based on the distortion ratio of the decoded image with respect to the first transform coefficient and the distortion ratio of the decoded image with respect to the second transform coefficient. According to an embodiment, the transform scheme determiner 1450 may select a transform coefficient corresponding to a decoded image having a smaller distortion rate as a transform coefficient for the target image.

The transformation scheme storage unit 1460 may store transformation scheme information on the selected transformation coefficient in a macroblock header for the target image. For example, when the transform coefficients are discrete sine transformed, the transform scheme storage unit 1460 may store the transform scheme information by displaying a flag bit of the macro block header as '1'. In addition, when the transform coefficient is discrete cosine transformed, the transform scheme storage unit 1460 may store the transform scheme information by displaying a flag bit of the macroblock header as '0'.

According to an embodiment, the transform scheme storage unit 1470 may store transform scheme information in consideration of an encoded block pattern of transform coefficients. The value of the coded block pattern indicates whether a transform coefficient exists. That is, when the transform coefficient is present, the value of the encoded block pattern may be '1', and when the transform coefficient is not present, the value of the encoded block pattern may be '0'. The transformation scheme storage unit 1470 may not store the transformation scheme information when the value of the encoded block pattern is '0'.

As described above, although the present invention has been described by way of limited embodiments and drawings, the present invention is not limited to the above embodiments, and those skilled in the art to which the present invention pertains various modifications and variations from such descriptions. This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined not only by the claims below but also by the equivalents of the claims.

Claims (3)

Determining whether a selective transform scheme has been applied to transform coefficients of the current block based on the first flag signaled from the image encoding apparatus; Here, the selective transformation technique is a transformation technique that selectively uses any one of a plurality of transformation techniques including discrete cosine transform (DCT) and discrete sine transform (DST).
When the selective transform scheme is applied to a transform coefficient of the current block according to the first flag, determining a transform scheme used by the current block based on a second flag of the current block; Here, the second flag specifies a transformation scheme used by the current block among the plurality of transformation schemes.
Generating a residual image by inversely transforming transform coefficients of the current block according to the determined transform scheme;
Generating a prediction image of the current block by performing prediction on the current block;
Generating a reconstructed image for the current block by adding the prediction image and the residual image; And
And applying a loop filter to the reconstructed image. Generating a prediction image of the current block by performing prediction on the current block;
Generating a residual image of the current block based on the subtraction of the prediction image from the current block;
Determining whether a selective transformation technique is to be applied to the coefficients of the residual image; Here, the selective transformation technique is a transformation technique that selectively uses any one of a plurality of transformation techniques including discrete cosine transform (DCT) and discrete sine transform (DST).
When the selective transformation technique is applied to the coefficients of the residual image, determining a transformation technique to be applied to the residual image among the plurality of transformation techniques; And
And transforming the coefficients of the residual image according to the determined transformation technique to generate transform coefficients of the current block. A computer-readable recording medium storing a bitstream received by a video decoding apparatus and decoded and used for reconstructing an image, wherein the bitstream includes:
A first flag of a current block and information about prediction of the current block,
The first flag is used to determine whether a selective transform scheme has been applied to the transform coefficients of the current block, wherein the selective transform scheme includes a plurality of discrete cosine transforms (DCTs) and discrete sine transforms (DSTs). Is a conversion technique that selectively uses any one of
When the selective transform scheme is applied to the transform coefficients of the current block according to the first flag, the second flag of the current block is used to determine a transform scheme used by the current block, wherein the second flag Specifies a transform scheme used by the current block among a plurality of transform schemes,
The determined transform technique is used to inversely transform the transform coefficients of the current block to generate a residual image.
The information about the prediction of the current block is used to generate a prediction image about the current block by performing prediction on the current block.
And storing a bitstream in which a reconstructed image for the current block is generated by adding the prediction image and the residual image.

Images